Spectrallic/MiniMax-H3-DMAD-int8-convrot
text-to-video model by Spectrallic
Older snapshot: check the original source for current information.
From the publisher
Model card & documentation
Source previewRead the publisher’s intended use, setup instructions, evaluations and limitations. The original model card is the source of truth.
These are the pruned MiniMax-H3 FL2VA and Ref2VA transformers from Comfy-Org/MiniMax-H3 with a full-rank DMAD LoRA merged in, quantized to INT8 ConvRot. The 4-step files use the fullcritic student and the 8-step files use 8stepfullcritic, which was trained for 8-step sampling. All four were made the same way and differ only in the base model and the LoRA: The tensor names and order, dtypes, shapes, data offsets and .comfyquant descriptors all match the Comfy-Org file. The difference is that the distillation is baked in, so you don't need a LoRA loader. Modified files: the weights differ from the MiniMax-H3 release, as described under How they were made. Layout: every block linear (blocks.N.attn.{qkvproj,outproj}, blocks.N.mlp.{fc1,fc2}) is INT8 and has its own .comfyquant tensor ({"format": "int8tensorwise", "convrot": true, "convrotgroupsize": 256}) and a per-row F32 weightscale. Everything else keeps the Comfy-Org dtype. Not trained on Ref2VA: DMAD was trained on the text-to-audio-video transformer, and…
Read the full model card ↗ · Preview checked 2026-11-10T07:07:24.542Z
Inside the original model card — Document outline
- MiniMax-H3 FL2VA and Ref2VA (pruned) 4-step and 8-step DMAD: INT8 ConvRot
- Examples
- FL2VA, 4 steps
- Ref2VA, 4 steps
- FL2VA, 8 steps
- Ref2VA, 8 steps
- Sampling
- How they were made
- Checks
- Credits & license
Headings are captured from the source. Links open the publisher’s document, not a locally hosted copy.
Links from the model card
References supplied by the publisher, not independently verified endorsements. Check the destination before downloading files or entering credentials.
Documentation belongs to its respective authors. Reported project/model license: other. A listing is not a grant of reuse or training rights. Confirm the document’s own terms at the source.
Model overview
Spectrallic/MiniMax-H3-DMAD-int8-convrot is a text-to-video model repository published by Spectrallic on Hugging Face. The source reports the diffusion-single-file library.
This page summarizes Hub metadata. For intended use, training data, evaluation results and limitations, consult the original model card.
Model facts
- Task
- text-to-video
- Library
- diffusion-single-file
- Recent downloads (30 days)
- 6,527
- Cumulative likes
- 3
- Hugging Face trending score
- 2
- Reported safetensors parameters
- Not reported
- Architecture
- Not reported
- License
- other
- Access
- Not gated by Hugging Face
- Created
- 2026-11-01T10:19:44.000Z
- Last modified
- 2026-11-04T14:20:24.000Z
Compatibility and lineage
Language coverage was not reported.
Reported base models: Comfy-Org/MiniMax-H3, ZhengmingYu/DMAD
Parameter count is not a RAM/VRAM requirement. Check precision, quantization, context length and runtime compatibility in the model card; no hardware or API-cost claim is inferred here.
Use and evaluate
Check license terms and access requirements first. Review the model card’s documented loading instructions, evaluations and safety limitations. Benchmark scores are not imported by this directory, and popularity is not an accuracy ranking.
No model weights are downloaded or executed by AltAPIs. Never enable remote model code without reviewing it.
Source and freshness
Source: Hugging Face Hub. Metadata observed 2026-11-11T06:57:46.557Z. Daily imports are snapshots, not real-time monitoring.
Popularity and source listings do not establish security, suitability, licensing rights or benchmark performance.
Related models
QuantStack/Wan2.2-T2V-A14B-GGUF
QuantStack/Wan2.2-TI2V-5B-GGUF